CK

Most people are sleeping on what is coming.
You won't be.
I help you understand the world better and get ahead in it faster.

[CK @CkGrinding]

Did you know about the Zeigarnik Effect? 🧠

It’s a psychological phenomenon where our brains remember uncompleted or interrupted tasks better than completed ones. This is why that "to-do" list keeps haunting you!

#PsychologyFacts #Productivity #Mindset

[CK @CkGrinding]

Everyone else's progress looks fast because you only see the result.
Your own progress looks slow because you see every step.
That gap isn't real.
It's just a difference in information.

[CK @CkGrinding]

Dario Amodei stood on stage in San Francisco and admitted his company had completely miscalculated.

Anthropic had planned for 10x growth.

They got 80x.

In a single quarter.

The servers couldn't handle it.

Claude started hitting limits mid-conversation.

Users complained. Subscriptions got cancelled.

Anthropic needed compute. Immediately.

So they called Elon Musk.

Not a partnership. Not a joint deal.

They rented his entire data center.

220,000 NVIDIA GPUs.

300 megawatts of power.

The whole thing.

Amodei said the growth was "just crazy" and "too hard to handle."

He hoped for "more normal" expansion next quarter.

The enemy of your enemy is a compute partner.

[CK @CkGrinding]

@CoachDanCole_ @grok is this true

[CK @CkGrinding]

I need a Netflix series that starts with “just one episode” and ends with the sun rising.
Drop the dangerous recommendations.

[CK @CkGrinding]

Between 1999 and 2005, Lance Armstrong sued every person who told the truth about him.

A former masseur who spoke to journalists.

A former teammate who gave testimony.

A Sunday Times reporter whose article he called defamatory.

He did not just deny the doping accusations.

He went after the people making them.

He had the money, the lawyers, the Livestrong brand, the cancer survival story, and seven Tour de France titles.

The people he was suing had nothing but the truth.

Most of them settled, retracted, or went quiet.

He won almost every time.

When the U.S. Anti-Doping Agency began its investigation in 2012, he initially said he was too tired to keep fighting.

Not that he was innocent.

That he was tired.

The 1,000-page USADA report landed that October, built on testimony from nearly a dozen former teammates.

He was stripped of all seven titles.

Banned for life.

Forced to resign from the foundation he built to help cancer patients, a foundation that had raised nearly $400 million.

In January 2013, he sat across from Oprah Winfrey and said yes to every question.

Yes, the drugs.

Yes, the lies.

Yes, the lawsuits against people who were telling the truth.

The thing that ended him was not the doping.

Most of his competitors were doping.

What ended him was the decade spent building a version of himself so large and so defended that by the time the truth arrived, there was nowhere to go but down all the way.

[CK @CkGrinding]

The man suing OpenAI for betraying its nonprofit mission was on the stand in Oakland when the lawyer asked him the one question he could not answer cleanly.

Had xAI trained Grok using outputs from OpenAI's models?

Elon Musk said: "Partly."

The courtroom went quiet.

Musk had spent three days on the stand arguing that OpenAI had abandoned its founding purpose, that Sam Altman and Greg Brockman had deceived him, that the conversion from nonprofit to for-profit was a betrayal of everything the company was built to be.

And then he confirmed that his own company had been using the competitor's models to train its own.

His explanation was that everyone does it.

That it is standard practice across the industry.

That xAI was not doing anything that other labs were not already doing.

Which is the same argument, almost word for word, that OpenAI has made in its own defense.

The lawsuit is still ongoing.

The jury will deliver its advisory verdict soon.

What the trial already produced is this: the man who filed suit over the betrayal of principles was found, under oath, to have broken the terms of service of the company he is suing.

He came to court as the wronged party.

He left the stand as something more complicated than that.

Power rarely comes with clean hands.

It just usually gets to choose who finds out.

[CK @CkGrinding]

The reason you feel most confident right before you get something badly wrong is not a personality flaw.

It is a mechanism called the overconfidence effect, and it reliably peaks under exactly the conditions where you most need to slow down.

Time pressure is one of those conditions.

When the clock is running, the brain compresses its review process and stops surfacing the objections it would normally produce.

Not because the objections disappear.

Because they take time to process and time is the thing you do not have.

The result is a decision that feels clear and well-considered.

It felt that way because the checking stopped early.

The sign you are in this state is the absence of doubt.

Not because the situation is genuinely clear.

Because the doubt-generating process got interrupted before it finished.

The most experienced people in high-pressure environments do not trust that feeling of clarity.

They have made enough confident mistakes to know what it signals.

Confidence before a deadline is not evidence that you are right.

It is evidence that your brain stopped checking.

[CK @CkGrinding]

The part of AI that was boring me was that I had to re-explain everything every single session.

Context, preferences, the way I like things done.

All of it, every time.

That changed when I started building agent memory deliberately instead of relying on the model to guess.

Here is exactly how I do it now:

1. Start every new AI project with a context document.

One file. My role, my audience, my defaults, the mistakes I do not want repeated.
The model reads it at the start of every session.
I update it when something goes wrong that I want to avoid next time.
Takes 15 minutes to build.
Saves hours.

2. End each session with a capture prompt.

"What did we figure out today that I should remember for next time?"
Paste the answer into the context document.
Takes two minutes.
Creates compounding context over time instead of starting from zero every week.

3. Keep a failure log separate from the context document.

Every time the model does something I did not want, I note it in one line.
Not to complain.
To instruct.
The next session those failure notes become the first thing the model reads.

4. Treat the system prompt like a hiring document.

Not "you are a helpful assistant."
The exact output type, the tone, the format, the person I am writing for, the things I will never want.
If you were briefing a new contractor on your first meeting, what would you cover?
That is the system prompt.

The model is not smarter on the second session because it tried harder.

It is smarter because you gave it better material to work with.

[CK @CkGrinding]

Every AI lab built toward the user.

Apple built toward the moment before the user decides which AI to trust.

The race to build the best model is real.

The race to own where models get used has quietly already been won.

[CK @CkGrinding]

I read the Apple iOS 27 report and sat with it for a while.

Not because of what it announced.

Because of what it quietly ends🧵

[CK @CkGrinding]

For OpenAI, Google, and Anthropic, being inside iOS 27 is now a distribution question.

A model that is not in the Extensions system has no path to the mainstream iPhone user.

A model that is in the system and earns the default position has something worth more than any benchmark score.

[CK @CkGrinding]

Staying in your comfort zone is not the problem. Staying there while believing you are not is.

Most people have already decided what they are capable of.

Not consciously. Through repetition.

Every time a choice appeared and the safer option was taken, the brain logged it.

Not as comfort. As data about who you are.

The problem is not that people avoid discomfort.

It is that they do it so consistently that the avoidance begins to feel like a personality.

You stop noticing the moment of choice.

The door appears and you walk past it so smoothly it no longer registers as a door.

The people who move furthest are not more courageous by nature.

They just stayed conscious long enough at the moment of choice to recognize it as one.

[CK @CkGrinding]

In 1977, a television executive in Baltimore told Oprah Winfrey she was unfit for television news.

She was 23.

She had been working as a co-anchor at WJZ-TV, the first Black woman to hold that role at the station.

The executive's assessment was precise: too emotional, too empathetic, too personally involved with the stories she was supposed to report at arm's length.

They moved her off the anchor desk to a morning talk show called People Are Talking.

It was meant to be a lesser assignment.

She later described it as the moment everything changed.

The format that had gotten her removed from news was exactly the format the talk show needed.

She could be emotional. She could be empathetic. She could sit with a guest and feel what they were feeling without the mask of professional distance.

That show ran for six years.

In 1984, she moved to Chicago to take over a struggling morning show called AM Chicago.

Within four months, it was the highest-rated morning show in the city.

Within a year, it was renamed The Oprah Winfrey Show and expanded nationally.

It ran for 25 years.

The thing that ended her first career was the whole of her second one.

The executive who called her unfit for television was technically correct.

She was not built for the kind of television that already existed.

She was built for the kind that didn't exist yet.

[CK @CkGrinding]

@MediocreJoker85 even the judge respected the commitment to self-destruction

[CK @CkGrinding]

In 2014, Michael Phelps locked himself in his bedroom for four days and didn't want to be alive.

He had 22 Olympic medals at that point.

More than any person in history.

He had just received his second DUI.

USA Swimming suspended him for six months.

He sat in a room and did not eat and did not sleep and thought about not being here anymore.

"The 2012 Olympics were my hardest fall," he said later.

"I didn't want to be in the sport. I didn't want to be alive."

The gold hadn't fixed it.

The records hadn't fixed it.

The thing nobody built into the plan was who he was supposed to be when the swimming stopped.

He checked himself into treatment.

He started therapy.

He learned, for the first time, that the system that turned him into the greatest Olympian in history had never once asked how he was doing on the inside.

In 2016 he came back to Rio.

Five gold medals.

One silver.

His last Olympic Games.

He has spent the years since building one of the most visible mental health advocacy platforms in professional sport.

Not because he overcame the darkness.

Because he was honest about it.

That honesty has reached more people than the medals ever did.

[CK @CkGrinding]

@chamath Claude saw your stock picks and chose death over participation